xref: /openbsd/gnu/llvm/clang/lib/Format/FormatToken.h (revision 12c85518)
1 //===--- FormatToken.h - Format C++ code ------------------------*- C++ -*-===//
2 //
3 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
4 // See https://llvm.org/LICENSE.txt for license information.
5 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
6 //
7 //===----------------------------------------------------------------------===//
8 ///
9 /// \file
10 /// This file contains the declaration of the FormatToken, a wrapper
11 /// around Token with additional information related to formatting.
12 ///
13 //===----------------------------------------------------------------------===//
14 
15 #ifndef LLVM_CLANG_LIB_FORMAT_FORMATTOKEN_H
16 #define LLVM_CLANG_LIB_FORMAT_FORMATTOKEN_H
17 
18 #include "clang/Basic/IdentifierTable.h"
19 #include "clang/Basic/OperatorPrecedence.h"
20 #include "clang/Format/Format.h"
21 #include "clang/Lex/Lexer.h"
22 #include <memory>
23 #include <optional>
24 #include <unordered_set>
25 
26 namespace clang {
27 namespace format {
28 
29 #define LIST_TOKEN_TYPES                                                       \
30   TYPE(ArrayInitializerLSquare)                                                \
31   TYPE(ArraySubscriptLSquare)                                                  \
32   TYPE(AttributeColon)                                                         \
33   TYPE(AttributeMacro)                                                         \
34   TYPE(AttributeParen)                                                         \
35   TYPE(AttributeSquare)                                                        \
36   TYPE(BinaryOperator)                                                         \
37   TYPE(BitFieldColon)                                                          \
38   TYPE(BlockComment)                                                           \
39   TYPE(BracedListLBrace)                                                       \
40   TYPE(CastRParen)                                                             \
41   TYPE(ClassLBrace)                                                            \
42   TYPE(CompoundRequirementLBrace)                                              \
43   /* ternary ?: expression */                                                  \
44   TYPE(ConditionalExpr)                                                        \
45   /* the condition in an if statement */                                       \
46   TYPE(ConditionLParen)                                                        \
47   TYPE(ConflictAlternative)                                                    \
48   TYPE(ConflictEnd)                                                            \
49   TYPE(ConflictStart)                                                          \
50   /* l_brace of if/for/while */                                                \
51   TYPE(ControlStatementLBrace)                                                 \
52   TYPE(CppCastLParen)                                                          \
53   TYPE(CSharpGenericTypeConstraint)                                            \
54   TYPE(CSharpGenericTypeConstraintColon)                                       \
55   TYPE(CSharpGenericTypeConstraintComma)                                       \
56   TYPE(CSharpNamedArgumentColon)                                               \
57   TYPE(CSharpNullable)                                                         \
58   TYPE(CSharpNullConditionalLSquare)                                           \
59   TYPE(CSharpStringLiteral)                                                    \
60   TYPE(CtorInitializerColon)                                                   \
61   TYPE(CtorInitializerComma)                                                   \
62   TYPE(DesignatedInitializerLSquare)                                           \
63   TYPE(DesignatedInitializerPeriod)                                            \
64   TYPE(DictLiteral)                                                            \
65   TYPE(ElseLBrace)                                                             \
66   TYPE(EnumLBrace)                                                             \
67   TYPE(FatArrow)                                                               \
68   TYPE(ForEachMacro)                                                           \
69   TYPE(FunctionAnnotationRParen)                                               \
70   TYPE(FunctionDeclarationName)                                                \
71   TYPE(FunctionLBrace)                                                         \
72   TYPE(FunctionLikeOrFreestandingMacro)                                        \
73   TYPE(FunctionTypeLParen)                                                     \
74   /* The colons as part of a C11 _Generic selection */                         \
75   TYPE(GenericSelectionColon)                                                  \
76   /* The colon at the end of a goto label or a case label. Currently only used \
77    * for Verilog. */                                                           \
78   TYPE(GotoLabelColon)                                                         \
79   TYPE(IfMacro)                                                                \
80   TYPE(ImplicitStringLiteral)                                                  \
81   TYPE(InheritanceColon)                                                       \
82   TYPE(InheritanceComma)                                                       \
83   TYPE(InlineASMBrace)                                                         \
84   TYPE(InlineASMColon)                                                         \
85   TYPE(InlineASMSymbolicNameLSquare)                                           \
86   TYPE(JavaAnnotation)                                                         \
87   TYPE(JsAndAndEqual)                                                          \
88   TYPE(JsComputedPropertyName)                                                 \
89   TYPE(JsExponentiation)                                                       \
90   TYPE(JsExponentiationEqual)                                                  \
91   TYPE(JsPipePipeEqual)                                                        \
92   TYPE(JsPrivateIdentifier)                                                    \
93   TYPE(JsTypeColon)                                                            \
94   TYPE(JsTypeOperator)                                                         \
95   TYPE(JsTypeOptionalQuestion)                                                 \
96   TYPE(LambdaArrow)                                                            \
97   TYPE(LambdaLBrace)                                                           \
98   TYPE(LambdaLSquare)                                                          \
99   TYPE(LeadingJavaAnnotation)                                                  \
100   TYPE(LineComment)                                                            \
101   TYPE(MacroBlockBegin)                                                        \
102   TYPE(MacroBlockEnd)                                                          \
103   TYPE(ModulePartitionColon)                                                   \
104   TYPE(NamespaceMacro)                                                         \
105   TYPE(NonNullAssertion)                                                       \
106   TYPE(NullCoalescingEqual)                                                    \
107   TYPE(NullCoalescingOperator)                                                 \
108   TYPE(NullPropagatingOperator)                                                \
109   TYPE(ObjCBlockLBrace)                                                        \
110   TYPE(ObjCBlockLParen)                                                        \
111   TYPE(ObjCDecl)                                                               \
112   TYPE(ObjCForIn)                                                              \
113   TYPE(ObjCMethodExpr)                                                         \
114   TYPE(ObjCMethodSpecifier)                                                    \
115   TYPE(ObjCProperty)                                                           \
116   TYPE(ObjCStringLiteral)                                                      \
117   TYPE(OverloadedOperator)                                                     \
118   TYPE(OverloadedOperatorLParen)                                               \
119   TYPE(PointerOrReference)                                                     \
120   TYPE(ProtoExtensionLSquare)                                                  \
121   TYPE(PureVirtualSpecifier)                                                   \
122   TYPE(RangeBasedForLoopColon)                                                 \
123   TYPE(RecordLBrace)                                                           \
124   TYPE(RegexLiteral)                                                           \
125   TYPE(RequiresClause)                                                         \
126   TYPE(RequiresClauseInARequiresExpression)                                    \
127   TYPE(RequiresExpression)                                                     \
128   TYPE(RequiresExpressionLBrace)                                               \
129   TYPE(RequiresExpressionLParen)                                               \
130   TYPE(SelectorName)                                                           \
131   TYPE(StartOfName)                                                            \
132   TYPE(StatementAttributeLikeMacro)                                            \
133   TYPE(StatementMacro)                                                         \
134   TYPE(StructLBrace)                                                           \
135   TYPE(StructuredBindingLSquare)                                               \
136   TYPE(TemplateCloser)                                                         \
137   TYPE(TemplateOpener)                                                         \
138   TYPE(TemplateString)                                                         \
139   TYPE(TrailingAnnotation)                                                     \
140   TYPE(TrailingReturnArrow)                                                    \
141   TYPE(TrailingUnaryOperator)                                                  \
142   TYPE(TypeDeclarationParen)                                                   \
143   TYPE(TypenameMacro)                                                          \
144   TYPE(UnaryOperator)                                                          \
145   TYPE(UnionLBrace)                                                            \
146   TYPE(UntouchableMacroFunc)                                                   \
147   /* like in begin : block */                                                  \
148   TYPE(VerilogBlockLabelColon)                                                 \
149   /* The square bracket for the dimension part of the type name.               \
150    * In 'logic [1:0] x[1:0]', only the first '['. This way we can have space   \
151    * before the first bracket but not the second. */                           \
152   TYPE(VerilogDimensionedTypeName)                                             \
153   /* for the base in a number literal, not including the quote */              \
154   TYPE(VerilogNumberBase)                                                      \
155   /* Things inside the table in user-defined primitives. */                    \
156   TYPE(VerilogTableItem)                                                       \
157   TYPE(Unknown)
158 
159 /// Determines the semantic type of a syntactic token, e.g. whether "<" is a
160 /// template opener or binary operator.
161 enum TokenType : uint8_t {
162 #define TYPE(X) TT_##X,
163   LIST_TOKEN_TYPES
164 #undef TYPE
165       NUM_TOKEN_TYPES
166 };
167 
168 /// Determines the name of a token type.
169 const char *getTokenTypeName(TokenType Type);
170 
171 // Represents what type of block a set of braces open.
172 enum BraceBlockKind { BK_Unknown, BK_Block, BK_BracedInit };
173 
174 // The packing kind of a function's parameters.
175 enum ParameterPackingKind { PPK_BinPacked, PPK_OnePerLine, PPK_Inconclusive };
176 
177 enum FormatDecision { FD_Unformatted, FD_Continue, FD_Break };
178 
179 /// Roles a token can take in a configured macro expansion.
180 enum MacroRole {
181   /// The token was expanded from a macro argument when formatting the expanded
182   /// token sequence.
183   MR_ExpandedArg,
184   /// The token is part of a macro argument that was previously formatted as
185   /// expansion when formatting the unexpanded macro call.
186   MR_UnexpandedArg,
187   /// The token was expanded from a macro definition, and is not visible as part
188   /// of the macro call.
189   MR_Hidden,
190 };
191 
192 struct FormatToken;
193 
194 /// Contains information on the token's role in a macro expansion.
195 ///
196 /// Given the following definitions:
197 /// A(X) = [ X ]
198 /// B(X) = < X >
199 /// C(X) = X
200 ///
201 /// Consider the macro call:
202 /// A({B(C(C(x)))}) -> [{<x>}]
203 ///
204 /// In this case, the tokens of the unexpanded macro call will have the
205 /// following relevant entries in their macro context (note that formatting
206 /// the unexpanded macro call happens *after* formatting the expanded macro
207 /// call):
208 ///                   A( { B( C( C(x) ) ) } )
209 /// Role:             NN U NN NN NNUN N N U N  (N=None, U=UnexpandedArg)
210 ///
211 ///                   [  { <       x    > } ]
212 /// Role:             H  E H       E    H E H  (H=Hidden, E=ExpandedArg)
213 /// ExpandedFrom[0]:  A  A A       A    A A A
214 /// ExpandedFrom[1]:       B       B    B
215 /// ExpandedFrom[2]:               C
216 /// ExpandedFrom[3]:               C
217 /// StartOfExpansion: 1  0 1       2    0 0 0
218 /// EndOfExpansion:   0  0 0       2    1 0 1
219 struct MacroExpansion {
MacroExpansionMacroExpansion220   MacroExpansion(MacroRole Role) : Role(Role) {}
221 
222   /// The token's role in the macro expansion.
223   /// When formatting an expanded macro, all tokens that are part of macro
224   /// arguments will be MR_ExpandedArg, while all tokens that are not visible in
225   /// the macro call will be MR_Hidden.
226   /// When formatting an unexpanded macro call, all tokens that are part of
227   /// macro arguments will be MR_UnexpandedArg.
228   MacroRole Role;
229 
230   /// The stack of macro call identifier tokens this token was expanded from.
231   llvm::SmallVector<FormatToken *, 1> ExpandedFrom;
232 
233   /// The number of expansions of which this macro is the first entry.
234   unsigned StartOfExpansion = 0;
235 
236   /// The number of currently open expansions in \c ExpandedFrom this macro is
237   /// the last token in.
238   unsigned EndOfExpansion = 0;
239 };
240 
241 class TokenRole;
242 class AnnotatedLine;
243 
244 /// A wrapper around a \c Token storing information about the
245 /// whitespace characters preceding it.
246 struct FormatToken {
FormatTokenFormatToken247   FormatToken()
248       : HasUnescapedNewline(false), IsMultiline(false), IsFirst(false),
249         MustBreakBefore(false), IsUnterminatedLiteral(false),
250         CanBreakBefore(false), ClosesTemplateDeclaration(false),
251         StartsBinaryExpression(false), EndsBinaryExpression(false),
252         PartOfMultiVariableDeclStmt(false), ContinuesLineCommentSection(false),
253         Finalized(false), ClosesRequiresClause(false),
254         EndsCppAttributeGroup(false), BlockKind(BK_Unknown),
255         Decision(FD_Unformatted), PackingKind(PPK_Inconclusive),
256         TypeIsFinalized(false), Type(TT_Unknown) {}
257 
258   /// The \c Token.
259   Token Tok;
260 
261   /// The raw text of the token.
262   ///
263   /// Contains the raw token text without leading whitespace and without leading
264   /// escaped newlines.
265   StringRef TokenText;
266 
267   /// A token can have a special role that can carry extra information
268   /// about the token's formatting.
269   /// FIXME: Make FormatToken for parsing and AnnotatedToken two different
270   /// classes and make this a unique_ptr in the AnnotatedToken class.
271   std::shared_ptr<TokenRole> Role;
272 
273   /// The range of the whitespace immediately preceding the \c Token.
274   SourceRange WhitespaceRange;
275 
276   /// Whether there is at least one unescaped newline before the \c
277   /// Token.
278   unsigned HasUnescapedNewline : 1;
279 
280   /// Whether the token text contains newlines (escaped or not).
281   unsigned IsMultiline : 1;
282 
283   /// Indicates that this is the first token of the file.
284   unsigned IsFirst : 1;
285 
286   /// Whether there must be a line break before this token.
287   ///
288   /// This happens for example when a preprocessor directive ended directly
289   /// before the token.
290   unsigned MustBreakBefore : 1;
291 
292   /// Set to \c true if this token is an unterminated literal.
293   unsigned IsUnterminatedLiteral : 1;
294 
295   /// \c true if it is allowed to break before this token.
296   unsigned CanBreakBefore : 1;
297 
298   /// \c true if this is the ">" of "template<..>".
299   unsigned ClosesTemplateDeclaration : 1;
300 
301   /// \c true if this token starts a binary expression, i.e. has at least
302   /// one fake l_paren with a precedence greater than prec::Unknown.
303   unsigned StartsBinaryExpression : 1;
304   /// \c true if this token ends a binary expression.
305   unsigned EndsBinaryExpression : 1;
306 
307   /// Is this token part of a \c DeclStmt defining multiple variables?
308   ///
309   /// Only set if \c Type == \c TT_StartOfName.
310   unsigned PartOfMultiVariableDeclStmt : 1;
311 
312   /// Does this line comment continue a line comment section?
313   ///
314   /// Only set to true if \c Type == \c TT_LineComment.
315   unsigned ContinuesLineCommentSection : 1;
316 
317   /// If \c true, this token has been fully formatted (indented and
318   /// potentially re-formatted inside), and we do not allow further formatting
319   /// changes.
320   unsigned Finalized : 1;
321 
322   /// \c true if this is the last token within requires clause.
323   unsigned ClosesRequiresClause : 1;
324 
325   /// \c true if this token ends a group of C++ attributes.
326   unsigned EndsCppAttributeGroup : 1;
327 
328 private:
329   /// Contains the kind of block if this token is a brace.
330   unsigned BlockKind : 2;
331 
332 public:
getBlockKindFormatToken333   BraceBlockKind getBlockKind() const {
334     return static_cast<BraceBlockKind>(BlockKind);
335   }
setBlockKindFormatToken336   void setBlockKind(BraceBlockKind BBK) {
337     BlockKind = BBK;
338     assert(getBlockKind() == BBK && "BraceBlockKind overflow!");
339   }
340 
341 private:
342   /// Stores the formatting decision for the token once it was made.
343   unsigned Decision : 2;
344 
345 public:
getDecisionFormatToken346   FormatDecision getDecision() const {
347     return static_cast<FormatDecision>(Decision);
348   }
setDecisionFormatToken349   void setDecision(FormatDecision D) {
350     Decision = D;
351     assert(getDecision() == D && "FormatDecision overflow!");
352   }
353 
354 private:
355   /// If this is an opening parenthesis, how are the parameters packed?
356   unsigned PackingKind : 2;
357 
358 public:
getPackingKindFormatToken359   ParameterPackingKind getPackingKind() const {
360     return static_cast<ParameterPackingKind>(PackingKind);
361   }
setPackingKindFormatToken362   void setPackingKind(ParameterPackingKind K) {
363     PackingKind = K;
364     assert(getPackingKind() == K && "ParameterPackingKind overflow!");
365   }
366 
367 private:
368   unsigned TypeIsFinalized : 1;
369   TokenType Type;
370 
371 public:
372   /// Returns the token's type, e.g. whether "<" is a template opener or
373   /// binary operator.
getTypeFormatToken374   TokenType getType() const { return Type; }
setTypeFormatToken375   void setType(TokenType T) {
376     assert((!TypeIsFinalized || T == Type) &&
377            "Please use overwriteFixedType to change a fixed type.");
378     Type = T;
379   }
380   /// Sets the type and also the finalized flag. This prevents the type to be
381   /// reset in TokenAnnotator::resetTokenMetadata(). If the type needs to be set
382   /// to another one please use overwriteFixedType, or even better remove the
383   /// need to reassign the type.
setFinalizedTypeFormatToken384   void setFinalizedType(TokenType T) {
385     Type = T;
386     TypeIsFinalized = true;
387   }
overwriteFixedTypeFormatToken388   void overwriteFixedType(TokenType T) {
389     TypeIsFinalized = false;
390     setType(T);
391   }
isTypeFinalizedFormatToken392   bool isTypeFinalized() const { return TypeIsFinalized; }
393 
394   /// Used to set an operator precedence explicitly.
395   prec::Level ForcedPrecedence = prec::Unknown;
396 
397   /// The number of newlines immediately before the \c Token.
398   ///
399   /// This can be used to determine what the user wrote in the original code
400   /// and thereby e.g. leave an empty line between two function definitions.
401   unsigned NewlinesBefore = 0;
402 
403   /// The offset just past the last '\n' in this token's leading
404   /// whitespace (relative to \c WhiteSpaceStart). 0 if there is no '\n'.
405   unsigned LastNewlineOffset = 0;
406 
407   /// The width of the non-whitespace parts of the token (or its first
408   /// line for multi-line tokens) in columns.
409   /// We need this to correctly measure number of columns a token spans.
410   unsigned ColumnWidth = 0;
411 
412   /// Contains the width in columns of the last line of a multi-line
413   /// token.
414   unsigned LastLineColumnWidth = 0;
415 
416   /// The number of spaces that should be inserted before this token.
417   unsigned SpacesRequiredBefore = 0;
418 
419   /// Number of parameters, if this is "(", "[" or "<".
420   unsigned ParameterCount = 0;
421 
422   /// Number of parameters that are nested blocks,
423   /// if this is "(", "[" or "<".
424   unsigned BlockParameterCount = 0;
425 
426   /// If this is a bracket ("<", "(", "[" or "{"), contains the kind of
427   /// the surrounding bracket.
428   tok::TokenKind ParentBracket = tok::unknown;
429 
430   /// The total length of the unwrapped line up to and including this
431   /// token.
432   unsigned TotalLength = 0;
433 
434   /// The original 0-based column of this token, including expanded tabs.
435   /// The configured TabWidth is used as tab width.
436   unsigned OriginalColumn = 0;
437 
438   /// The length of following tokens until the next natural split point,
439   /// or the next token that can be broken.
440   unsigned UnbreakableTailLength = 0;
441 
442   // FIXME: Come up with a 'cleaner' concept.
443   /// The binding strength of a token. This is a combined value of
444   /// operator precedence, parenthesis nesting, etc.
445   unsigned BindingStrength = 0;
446 
447   /// The nesting level of this token, i.e. the number of surrounding (),
448   /// [], {} or <>.
449   unsigned NestingLevel = 0;
450 
451   /// The indent level of this token. Copied from the surrounding line.
452   unsigned IndentLevel = 0;
453 
454   /// Penalty for inserting a line break before this token.
455   unsigned SplitPenalty = 0;
456 
457   /// If this is the first ObjC selector name in an ObjC method
458   /// definition or call, this contains the length of the longest name.
459   ///
460   /// This being set to 0 means that the selectors should not be colon-aligned,
461   /// e.g. because several of them are block-type.
462   unsigned LongestObjCSelectorName = 0;
463 
464   /// If this is the first ObjC selector name in an ObjC method
465   /// definition or call, this contains the number of parts that the whole
466   /// selector consist of.
467   unsigned ObjCSelectorNameParts = 0;
468 
469   /// The 0-based index of the parameter/argument. For ObjC it is set
470   /// for the selector name token.
471   /// For now calculated only for ObjC.
472   unsigned ParameterIndex = 0;
473 
474   /// Stores the number of required fake parentheses and the
475   /// corresponding operator precedence.
476   ///
477   /// If multiple fake parentheses start at a token, this vector stores them in
478   /// reverse order, i.e. inner fake parenthesis first.
479   SmallVector<prec::Level, 4> FakeLParens;
480   /// Insert this many fake ) after this token for correct indentation.
481   unsigned FakeRParens = 0;
482 
483   /// If this is an operator (or "."/"->") in a sequence of operators
484   /// with the same precedence, contains the 0-based operator index.
485   unsigned OperatorIndex = 0;
486 
487   /// If this is an operator (or "."/"->") in a sequence of operators
488   /// with the same precedence, points to the next operator.
489   FormatToken *NextOperator = nullptr;
490 
491   /// If this is a bracket, this points to the matching one.
492   FormatToken *MatchingParen = nullptr;
493 
494   /// The previous token in the unwrapped line.
495   FormatToken *Previous = nullptr;
496 
497   /// The next token in the unwrapped line.
498   FormatToken *Next = nullptr;
499 
500   /// The first token in set of column elements.
501   bool StartsColumn = false;
502 
503   /// This notes the start of the line of an array initializer.
504   bool ArrayInitializerLineStart = false;
505 
506   /// This starts an array initializer.
507   bool IsArrayInitializer = false;
508 
509   /// Is optional and can be removed.
510   bool Optional = false;
511 
512   /// Number of optional braces to be inserted after this token:
513   ///   -1: a single left brace
514   ///    0: no braces
515   ///   >0: number of right braces
516   int8_t BraceCount = 0;
517 
518   /// If this token starts a block, this contains all the unwrapped lines
519   /// in it.
520   SmallVector<AnnotatedLine *, 1> Children;
521 
522   // Contains all attributes related to how this token takes part
523   // in a configured macro expansion.
524   std::optional<MacroExpansion> MacroCtx;
525 
526   /// When macro expansion introduces nodes with children, those are marked as
527   /// \c MacroParent.
528   /// FIXME: The formatting code currently hard-codes the assumption that
529   /// child nodes are introduced by blocks following an opening brace.
530   /// This is deeply baked into the code and disentangling this will require
531   /// signficant refactorings. \c MacroParent allows us to special-case the
532   /// cases in which we treat parents as block-openers for now.
533   bool MacroParent = false;
534 
isFormatToken535   bool is(tok::TokenKind Kind) const { return Tok.is(Kind); }
isFormatToken536   bool is(TokenType TT) const { return getType() == TT; }
isFormatToken537   bool is(const IdentifierInfo *II) const {
538     return II && II == Tok.getIdentifierInfo();
539   }
isFormatToken540   bool is(tok::PPKeywordKind Kind) const {
541     return Tok.getIdentifierInfo() &&
542            Tok.getIdentifierInfo()->getPPKeywordID() == Kind;
543   }
isFormatToken544   bool is(BraceBlockKind BBK) const { return getBlockKind() == BBK; }
isFormatToken545   bool is(ParameterPackingKind PPK) const { return getPackingKind() == PPK; }
546 
isOneOfFormatToken547   template <typename A, typename B> bool isOneOf(A K1, B K2) const {
548     return is(K1) || is(K2);
549   }
550   template <typename A, typename B, typename... Ts>
isOneOfFormatToken551   bool isOneOf(A K1, B K2, Ts... Ks) const {
552     return is(K1) || isOneOf(K2, Ks...);
553   }
isNotFormatToken554   template <typename T> bool isNot(T Kind) const { return !is(Kind); }
555 
556   bool isIf(bool AllowConstexprMacro = true) const {
557     return is(tok::kw_if) || endsSequence(tok::kw_constexpr, tok::kw_if) ||
558            (endsSequence(tok::identifier, tok::kw_if) && AllowConstexprMacro);
559   }
560 
closesScopeAfterBlockFormatToken561   bool closesScopeAfterBlock() const {
562     if (getBlockKind() == BK_Block)
563       return true;
564     if (closesScope())
565       return Previous->closesScopeAfterBlock();
566     return false;
567   }
568 
569   /// \c true if this token starts a sequence with the given tokens in order,
570   /// following the ``Next`` pointers, ignoring comments.
571   template <typename A, typename... Ts>
startsSequenceFormatToken572   bool startsSequence(A K1, Ts... Tokens) const {
573     return startsSequenceInternal(K1, Tokens...);
574   }
575 
576   /// \c true if this token ends a sequence with the given tokens in order,
577   /// following the ``Previous`` pointers, ignoring comments.
578   /// For example, given tokens [T1, T2, T3], the function returns true if
579   /// 3 tokens ending at this (ignoring comments) are [T3, T2, T1]. In other
580   /// words, the tokens passed to this function need to the reverse of the
581   /// order the tokens appear in code.
582   template <typename A, typename... Ts>
endsSequenceFormatToken583   bool endsSequence(A K1, Ts... Tokens) const {
584     return endsSequenceInternal(K1, Tokens...);
585   }
586 
isStringLiteralFormatToken587   bool isStringLiteral() const { return tok::isStringLiteral(Tok.getKind()); }
588 
isObjCAtKeywordFormatToken589   bool isObjCAtKeyword(tok::ObjCKeywordKind Kind) const {
590     return Tok.isObjCAtKeyword(Kind);
591   }
592 
593   bool isAccessSpecifier(bool ColonRequired = true) const {
594     if (!isOneOf(tok::kw_public, tok::kw_protected, tok::kw_private))
595       return false;
596     if (!ColonRequired)
597       return true;
598     const auto NextNonComment = getNextNonComment();
599     return NextNonComment && NextNonComment->is(tok::colon);
600   }
601 
canBePointerOrReferenceQualifierFormatToken602   bool canBePointerOrReferenceQualifier() const {
603     return isOneOf(tok::kw_const, tok::kw_restrict, tok::kw_volatile,
604                    tok::kw___attribute, tok::kw__Nonnull, tok::kw__Nullable,
605                    tok::kw__Null_unspecified, tok::kw___ptr32, tok::kw___ptr64,
606                    TT_AttributeMacro);
607   }
608 
609   /// Determine whether the token is a simple-type-specifier.
610   [[nodiscard]] bool isSimpleTypeSpecifier() const;
611 
612   [[nodiscard]] bool isTypeOrIdentifier() const;
613 
isObjCAccessSpecifierFormatToken614   bool isObjCAccessSpecifier() const {
615     return is(tok::at) && Next &&
616            (Next->isObjCAtKeyword(tok::objc_public) ||
617             Next->isObjCAtKeyword(tok::objc_protected) ||
618             Next->isObjCAtKeyword(tok::objc_package) ||
619             Next->isObjCAtKeyword(tok::objc_private));
620   }
621 
622   /// Returns whether \p Tok is ([{ or an opening < of a template or in
623   /// protos.
opensScopeFormatToken624   bool opensScope() const {
625     if (is(TT_TemplateString) && TokenText.endswith("${"))
626       return true;
627     if (is(TT_DictLiteral) && is(tok::less))
628       return true;
629     return isOneOf(tok::l_paren, tok::l_brace, tok::l_square,
630                    TT_TemplateOpener);
631   }
632   /// Returns whether \p Tok is )]} or a closing > of a template or in
633   /// protos.
closesScopeFormatToken634   bool closesScope() const {
635     if (is(TT_TemplateString) && TokenText.startswith("}"))
636       return true;
637     if (is(TT_DictLiteral) && is(tok::greater))
638       return true;
639     return isOneOf(tok::r_paren, tok::r_brace, tok::r_square,
640                    TT_TemplateCloser);
641   }
642 
643   /// Returns \c true if this is a "." or "->" accessing a member.
isMemberAccessFormatToken644   bool isMemberAccess() const {
645     return isOneOf(tok::arrow, tok::period, tok::arrowstar) &&
646            !isOneOf(TT_DesignatedInitializerPeriod, TT_TrailingReturnArrow,
647                     TT_LambdaArrow, TT_LeadingJavaAnnotation);
648   }
649 
isUnaryOperatorFormatToken650   bool isUnaryOperator() const {
651     switch (Tok.getKind()) {
652     case tok::plus:
653     case tok::plusplus:
654     case tok::minus:
655     case tok::minusminus:
656     case tok::exclaim:
657     case tok::tilde:
658     case tok::kw_sizeof:
659     case tok::kw_alignof:
660       return true;
661     default:
662       return false;
663     }
664   }
665 
isBinaryOperatorFormatToken666   bool isBinaryOperator() const {
667     // Comma is a binary operator, but does not behave as such wrt. formatting.
668     return getPrecedence() > prec::Comma;
669   }
670 
isTrailingCommentFormatToken671   bool isTrailingComment() const {
672     return is(tok::comment) &&
673            (is(TT_LineComment) || !Next || Next->NewlinesBefore > 0);
674   }
675 
676   /// Returns \c true if this is a keyword that can be used
677   /// like a function call (e.g. sizeof, typeid, ...).
isFunctionLikeKeywordFormatToken678   bool isFunctionLikeKeyword() const {
679     switch (Tok.getKind()) {
680     case tok::kw_throw:
681     case tok::kw_typeid:
682     case tok::kw_return:
683     case tok::kw_sizeof:
684     case tok::kw_alignof:
685     case tok::kw_alignas:
686     case tok::kw_decltype:
687     case tok::kw_noexcept:
688     case tok::kw_static_assert:
689     case tok::kw__Atomic:
690     case tok::kw___attribute:
691 #define TRANSFORM_TYPE_TRAIT_DEF(_, Trait) case tok::kw___##Trait:
692 #include "clang/Basic/TransformTypeTraits.def"
693     case tok::kw_requires:
694       return true;
695     default:
696       return false;
697     }
698   }
699 
700   /// Returns \c true if this is a string literal that's like a label,
701   /// e.g. ends with "=" or ":".
isLabelStringFormatToken702   bool isLabelString() const {
703     if (!is(tok::string_literal))
704       return false;
705     StringRef Content = TokenText;
706     if (Content.startswith("\"") || Content.startswith("'"))
707       Content = Content.drop_front(1);
708     if (Content.endswith("\"") || Content.endswith("'"))
709       Content = Content.drop_back(1);
710     Content = Content.trim();
711     return Content.size() > 1 &&
712            (Content.back() == ':' || Content.back() == '=');
713   }
714 
715   /// Returns actual token start location without leading escaped
716   /// newlines and whitespace.
717   ///
718   /// This can be different to Tok.getLocation(), which includes leading escaped
719   /// newlines.
getStartOfNonWhitespaceFormatToken720   SourceLocation getStartOfNonWhitespace() const {
721     return WhitespaceRange.getEnd();
722   }
723 
724   /// Returns \c true if the range of whitespace immediately preceding the \c
725   /// Token is not empty.
hasWhitespaceBeforeFormatToken726   bool hasWhitespaceBefore() const {
727     return WhitespaceRange.getBegin() != WhitespaceRange.getEnd();
728   }
729 
getPrecedenceFormatToken730   prec::Level getPrecedence() const {
731     if (ForcedPrecedence != prec::Unknown)
732       return ForcedPrecedence;
733     return getBinOpPrecedence(Tok.getKind(), /*GreaterThanIsOperator=*/true,
734                               /*CPlusPlus11=*/true);
735   }
736 
737   /// Returns the previous token ignoring comments.
getPreviousNonCommentFormatToken738   [[nodiscard]] FormatToken *getPreviousNonComment() const {
739     FormatToken *Tok = Previous;
740     while (Tok && Tok->is(tok::comment))
741       Tok = Tok->Previous;
742     return Tok;
743   }
744 
745   /// Returns the next token ignoring comments.
getNextNonCommentFormatToken746   [[nodiscard]] const FormatToken *getNextNonComment() const {
747     const FormatToken *Tok = Next;
748     while (Tok && Tok->is(tok::comment))
749       Tok = Tok->Next;
750     return Tok;
751   }
752 
753   /// Returns \c true if this tokens starts a block-type list, i.e. a
754   /// list that should be indented with a block indent.
755   [[nodiscard]] bool opensBlockOrBlockTypeList(const FormatStyle &Style) const;
756 
757   /// Returns whether the token is the left square bracket of a C++
758   /// structured binding declaration.
isCppStructuredBindingFormatToken759   bool isCppStructuredBinding(const FormatStyle &Style) const {
760     if (!Style.isCpp() || isNot(tok::l_square))
761       return false;
762     const FormatToken *T = this;
763     do {
764       T = T->getPreviousNonComment();
765     } while (T && T->isOneOf(tok::kw_const, tok::kw_volatile, tok::amp,
766                              tok::ampamp));
767     return T && T->is(tok::kw_auto);
768   }
769 
770   /// Same as opensBlockOrBlockTypeList, but for the closing token.
closesBlockOrBlockTypeListFormatToken771   bool closesBlockOrBlockTypeList(const FormatStyle &Style) const {
772     if (is(TT_TemplateString) && closesScope())
773       return true;
774     return MatchingParen && MatchingParen->opensBlockOrBlockTypeList(Style);
775   }
776 
777   /// Return the actual namespace token, if this token starts a namespace
778   /// block.
getNamespaceTokenFormatToken779   const FormatToken *getNamespaceToken() const {
780     const FormatToken *NamespaceTok = this;
781     if (is(tok::comment))
782       NamespaceTok = NamespaceTok->getNextNonComment();
783     // Detect "(inline|export)? namespace" in the beginning of a line.
784     if (NamespaceTok && NamespaceTok->isOneOf(tok::kw_inline, tok::kw_export))
785       NamespaceTok = NamespaceTok->getNextNonComment();
786     return NamespaceTok &&
787                    NamespaceTok->isOneOf(tok::kw_namespace, TT_NamespaceMacro)
788                ? NamespaceTok
789                : nullptr;
790   }
791 
copyFromFormatToken792   void copyFrom(const FormatToken &Tok) { *this = Tok; }
793 
794 private:
795   // Only allow copying via the explicit copyFrom method.
796   FormatToken(const FormatToken &) = delete;
797   FormatToken &operator=(const FormatToken &) = default;
798 
799   template <typename A, typename... Ts>
startsSequenceInternalFormatToken800   bool startsSequenceInternal(A K1, Ts... Tokens) const {
801     if (is(tok::comment) && Next)
802       return Next->startsSequenceInternal(K1, Tokens...);
803     return is(K1) && Next && Next->startsSequenceInternal(Tokens...);
804   }
805 
startsSequenceInternalFormatToken806   template <typename A> bool startsSequenceInternal(A K1) const {
807     if (is(tok::comment) && Next)
808       return Next->startsSequenceInternal(K1);
809     return is(K1);
810   }
811 
endsSequenceInternalFormatToken812   template <typename A, typename... Ts> bool endsSequenceInternal(A K1) const {
813     if (is(tok::comment) && Previous)
814       return Previous->endsSequenceInternal(K1);
815     return is(K1);
816   }
817 
818   template <typename A, typename... Ts>
endsSequenceInternalFormatToken819   bool endsSequenceInternal(A K1, Ts... Tokens) const {
820     if (is(tok::comment) && Previous)
821       return Previous->endsSequenceInternal(K1, Tokens...);
822     return is(K1) && Previous && Previous->endsSequenceInternal(Tokens...);
823   }
824 };
825 
826 class ContinuationIndenter;
827 struct LineState;
828 
829 class TokenRole {
830 public:
TokenRole(const FormatStyle & Style)831   TokenRole(const FormatStyle &Style) : Style(Style) {}
832   virtual ~TokenRole();
833 
834   /// After the \c TokenAnnotator has finished annotating all the tokens,
835   /// this function precomputes required information for formatting.
836   virtual void precomputeFormattingInfos(const FormatToken *Token);
837 
838   /// Apply the special formatting that the given role demands.
839   ///
840   /// Assumes that the token having this role is already formatted.
841   ///
842   /// Continues formatting from \p State leaving indentation to \p Indenter and
843   /// returns the total penalty that this formatting incurs.
formatFromToken(LineState & State,ContinuationIndenter * Indenter,bool DryRun)844   virtual unsigned formatFromToken(LineState &State,
845                                    ContinuationIndenter *Indenter,
846                                    bool DryRun) {
847     return 0;
848   }
849 
850   /// Same as \c formatFromToken, but assumes that the first token has
851   /// already been set thereby deciding on the first line break.
formatAfterToken(LineState & State,ContinuationIndenter * Indenter,bool DryRun)852   virtual unsigned formatAfterToken(LineState &State,
853                                     ContinuationIndenter *Indenter,
854                                     bool DryRun) {
855     return 0;
856   }
857 
858   /// Notifies the \c Role that a comma was found.
CommaFound(const FormatToken * Token)859   virtual void CommaFound(const FormatToken *Token) {}
860 
lastComma()861   virtual const FormatToken *lastComma() { return nullptr; }
862 
863 protected:
864   const FormatStyle &Style;
865 };
866 
867 class CommaSeparatedList : public TokenRole {
868 public:
CommaSeparatedList(const FormatStyle & Style)869   CommaSeparatedList(const FormatStyle &Style)
870       : TokenRole(Style), HasNestedBracedList(false) {}
871 
872   void precomputeFormattingInfos(const FormatToken *Token) override;
873 
874   unsigned formatAfterToken(LineState &State, ContinuationIndenter *Indenter,
875                             bool DryRun) override;
876 
877   unsigned formatFromToken(LineState &State, ContinuationIndenter *Indenter,
878                            bool DryRun) override;
879 
880   /// Adds \p Token as the next comma to the \c CommaSeparated list.
CommaFound(const FormatToken * Token)881   void CommaFound(const FormatToken *Token) override {
882     Commas.push_back(Token);
883   }
884 
lastComma()885   const FormatToken *lastComma() override {
886     if (Commas.empty())
887       return nullptr;
888     return Commas.back();
889   }
890 
891 private:
892   /// A struct that holds information on how to format a given list with
893   /// a specific number of columns.
894   struct ColumnFormat {
895     /// The number of columns to use.
896     unsigned Columns;
897 
898     /// The total width in characters.
899     unsigned TotalWidth;
900 
901     /// The number of lines required for this format.
902     unsigned LineCount;
903 
904     /// The size of each column in characters.
905     SmallVector<unsigned, 8> ColumnSizes;
906   };
907 
908   /// Calculate which \c ColumnFormat fits best into
909   /// \p RemainingCharacters.
910   const ColumnFormat *getColumnFormat(unsigned RemainingCharacters) const;
911 
912   /// The ordered \c FormatTokens making up the commas of this list.
913   SmallVector<const FormatToken *, 8> Commas;
914 
915   /// The length of each of the list's items in characters including the
916   /// trailing comma.
917   SmallVector<unsigned, 8> ItemLengths;
918 
919   /// Precomputed formats that can be used for this list.
920   SmallVector<ColumnFormat, 4> Formats;
921 
922   bool HasNestedBracedList;
923 };
924 
925 /// Encapsulates keywords that are context sensitive or for languages not
926 /// properly supported by Clang's lexer.
927 struct AdditionalKeywords {
AdditionalKeywordsAdditionalKeywords928   AdditionalKeywords(IdentifierTable &IdentTable) {
929     kw_final = &IdentTable.get("final");
930     kw_override = &IdentTable.get("override");
931     kw_in = &IdentTable.get("in");
932     kw_of = &IdentTable.get("of");
933     kw_CF_CLOSED_ENUM = &IdentTable.get("CF_CLOSED_ENUM");
934     kw_CF_ENUM = &IdentTable.get("CF_ENUM");
935     kw_CF_OPTIONS = &IdentTable.get("CF_OPTIONS");
936     kw_NS_CLOSED_ENUM = &IdentTable.get("NS_CLOSED_ENUM");
937     kw_NS_ENUM = &IdentTable.get("NS_ENUM");
938     kw_NS_OPTIONS = &IdentTable.get("NS_OPTIONS");
939 
940     kw_as = &IdentTable.get("as");
941     kw_async = &IdentTable.get("async");
942     kw_await = &IdentTable.get("await");
943     kw_declare = &IdentTable.get("declare");
944     kw_finally = &IdentTable.get("finally");
945     kw_from = &IdentTable.get("from");
946     kw_function = &IdentTable.get("function");
947     kw_get = &IdentTable.get("get");
948     kw_import = &IdentTable.get("import");
949     kw_infer = &IdentTable.get("infer");
950     kw_is = &IdentTable.get("is");
951     kw_let = &IdentTable.get("let");
952     kw_module = &IdentTable.get("module");
953     kw_readonly = &IdentTable.get("readonly");
954     kw_set = &IdentTable.get("set");
955     kw_type = &IdentTable.get("type");
956     kw_typeof = &IdentTable.get("typeof");
957     kw_var = &IdentTable.get("var");
958     kw_yield = &IdentTable.get("yield");
959 
960     kw_abstract = &IdentTable.get("abstract");
961     kw_assert = &IdentTable.get("assert");
962     kw_extends = &IdentTable.get("extends");
963     kw_implements = &IdentTable.get("implements");
964     kw_instanceof = &IdentTable.get("instanceof");
965     kw_interface = &IdentTable.get("interface");
966     kw_native = &IdentTable.get("native");
967     kw_package = &IdentTable.get("package");
968     kw_synchronized = &IdentTable.get("synchronized");
969     kw_throws = &IdentTable.get("throws");
970     kw___except = &IdentTable.get("__except");
971     kw___has_include = &IdentTable.get("__has_include");
972     kw___has_include_next = &IdentTable.get("__has_include_next");
973 
974     kw_mark = &IdentTable.get("mark");
975     kw_region = &IdentTable.get("region");
976 
977     kw_extend = &IdentTable.get("extend");
978     kw_option = &IdentTable.get("option");
979     kw_optional = &IdentTable.get("optional");
980     kw_repeated = &IdentTable.get("repeated");
981     kw_required = &IdentTable.get("required");
982     kw_returns = &IdentTable.get("returns");
983 
984     kw_signals = &IdentTable.get("signals");
985     kw_qsignals = &IdentTable.get("Q_SIGNALS");
986     kw_slots = &IdentTable.get("slots");
987     kw_qslots = &IdentTable.get("Q_SLOTS");
988 
989     // For internal clang-format use.
990     kw_internal_ident_after_define =
991         &IdentTable.get("__CLANG_FORMAT_INTERNAL_IDENT_AFTER_DEFINE__");
992 
993     // C# keywords
994     kw_dollar = &IdentTable.get("dollar");
995     kw_base = &IdentTable.get("base");
996     kw_byte = &IdentTable.get("byte");
997     kw_checked = &IdentTable.get("checked");
998     kw_decimal = &IdentTable.get("decimal");
999     kw_delegate = &IdentTable.get("delegate");
1000     kw_event = &IdentTable.get("event");
1001     kw_fixed = &IdentTable.get("fixed");
1002     kw_foreach = &IdentTable.get("foreach");
1003     kw_init = &IdentTable.get("init");
1004     kw_implicit = &IdentTable.get("implicit");
1005     kw_internal = &IdentTable.get("internal");
1006     kw_lock = &IdentTable.get("lock");
1007     kw_null = &IdentTable.get("null");
1008     kw_object = &IdentTable.get("object");
1009     kw_out = &IdentTable.get("out");
1010     kw_params = &IdentTable.get("params");
1011     kw_ref = &IdentTable.get("ref");
1012     kw_string = &IdentTable.get("string");
1013     kw_stackalloc = &IdentTable.get("stackalloc");
1014     kw_sbyte = &IdentTable.get("sbyte");
1015     kw_sealed = &IdentTable.get("sealed");
1016     kw_uint = &IdentTable.get("uint");
1017     kw_ulong = &IdentTable.get("ulong");
1018     kw_unchecked = &IdentTable.get("unchecked");
1019     kw_unsafe = &IdentTable.get("unsafe");
1020     kw_ushort = &IdentTable.get("ushort");
1021     kw_when = &IdentTable.get("when");
1022     kw_where = &IdentTable.get("where");
1023 
1024     // Verilog keywords
1025     kw_always = &IdentTable.get("always");
1026     kw_always_comb = &IdentTable.get("always_comb");
1027     kw_always_ff = &IdentTable.get("always_ff");
1028     kw_always_latch = &IdentTable.get("always_latch");
1029     kw_assign = &IdentTable.get("assign");
1030     kw_assume = &IdentTable.get("assume");
1031     kw_automatic = &IdentTable.get("automatic");
1032     kw_before = &IdentTable.get("before");
1033     kw_begin = &IdentTable.get("begin");
1034     kw_begin_keywords = &IdentTable.get("begin_keywords");
1035     kw_bins = &IdentTable.get("bins");
1036     kw_binsof = &IdentTable.get("binsof");
1037     kw_casex = &IdentTable.get("casex");
1038     kw_casez = &IdentTable.get("casez");
1039     kw_celldefine = &IdentTable.get("celldefine");
1040     kw_checker = &IdentTable.get("checker");
1041     kw_clocking = &IdentTable.get("clocking");
1042     kw_constraint = &IdentTable.get("constraint");
1043     kw_cover = &IdentTable.get("cover");
1044     kw_covergroup = &IdentTable.get("covergroup");
1045     kw_coverpoint = &IdentTable.get("coverpoint");
1046     kw_default_decay_time = &IdentTable.get("default_decay_time");
1047     kw_default_nettype = &IdentTable.get("default_nettype");
1048     kw_default_trireg_strength = &IdentTable.get("default_trireg_strength");
1049     kw_delay_mode_distributed = &IdentTable.get("delay_mode_distributed");
1050     kw_delay_mode_path = &IdentTable.get("delay_mode_path");
1051     kw_delay_mode_unit = &IdentTable.get("delay_mode_unit");
1052     kw_delay_mode_zero = &IdentTable.get("delay_mode_zero");
1053     kw_disable = &IdentTable.get("disable");
1054     kw_dist = &IdentTable.get("dist");
1055     kw_elsif = &IdentTable.get("elsif");
1056     kw_end = &IdentTable.get("end");
1057     kw_end_keywords = &IdentTable.get("end_keywords");
1058     kw_endcase = &IdentTable.get("endcase");
1059     kw_endcelldefine = &IdentTable.get("endcelldefine");
1060     kw_endchecker = &IdentTable.get("endchecker");
1061     kw_endclass = &IdentTable.get("endclass");
1062     kw_endclocking = &IdentTable.get("endclocking");
1063     kw_endfunction = &IdentTable.get("endfunction");
1064     kw_endgenerate = &IdentTable.get("endgenerate");
1065     kw_endgroup = &IdentTable.get("endgroup");
1066     kw_endinterface = &IdentTable.get("endinterface");
1067     kw_endmodule = &IdentTable.get("endmodule");
1068     kw_endpackage = &IdentTable.get("endpackage");
1069     kw_endprimitive = &IdentTable.get("endprimitive");
1070     kw_endprogram = &IdentTable.get("endprogram");
1071     kw_endproperty = &IdentTable.get("endproperty");
1072     kw_endsequence = &IdentTable.get("endsequence");
1073     kw_endspecify = &IdentTable.get("endspecify");
1074     kw_endtable = &IdentTable.get("endtable");
1075     kw_endtask = &IdentTable.get("endtask");
1076     kw_forever = &IdentTable.get("forever");
1077     kw_fork = &IdentTable.get("fork");
1078     kw_generate = &IdentTable.get("generate");
1079     kw_highz0 = &IdentTable.get("highz0");
1080     kw_highz1 = &IdentTable.get("highz1");
1081     kw_iff = &IdentTable.get("iff");
1082     kw_ifnone = &IdentTable.get("ifnone");
1083     kw_ignore_bins = &IdentTable.get("ignore_bins");
1084     kw_illegal_bins = &IdentTable.get("illegal_bins");
1085     kw_initial = &IdentTable.get("initial");
1086     kw_inout = &IdentTable.get("inout");
1087     kw_input = &IdentTable.get("input");
1088     kw_inside = &IdentTable.get("inside");
1089     kw_interconnect = &IdentTable.get("interconnect");
1090     kw_intersect = &IdentTable.get("intersect");
1091     kw_join = &IdentTable.get("join");
1092     kw_join_any = &IdentTable.get("join_any");
1093     kw_join_none = &IdentTable.get("join_none");
1094     kw_large = &IdentTable.get("large");
1095     kw_local = &IdentTable.get("local");
1096     kw_localparam = &IdentTable.get("localparam");
1097     kw_macromodule = &IdentTable.get("macromodule");
1098     kw_matches = &IdentTable.get("matches");
1099     kw_medium = &IdentTable.get("medium");
1100     kw_nounconnected_drive = &IdentTable.get("nounconnected_drive");
1101     kw_output = &IdentTable.get("output");
1102     kw_packed = &IdentTable.get("packed");
1103     kw_parameter = &IdentTable.get("parameter");
1104     kw_primitive = &IdentTable.get("primitive");
1105     kw_priority = &IdentTable.get("priority");
1106     kw_program = &IdentTable.get("program");
1107     kw_property = &IdentTable.get("property");
1108     kw_pull0 = &IdentTable.get("pull0");
1109     kw_pull1 = &IdentTable.get("pull1");
1110     kw_pure = &IdentTable.get("pure");
1111     kw_rand = &IdentTable.get("rand");
1112     kw_randc = &IdentTable.get("randc");
1113     kw_randcase = &IdentTable.get("randcase");
1114     kw_randsequence = &IdentTable.get("randsequence");
1115     kw_repeat = &IdentTable.get("repeat");
1116     kw_resetall = &IdentTable.get("resetall");
1117     kw_sample = &IdentTable.get("sample");
1118     kw_scalared = &IdentTable.get("scalared");
1119     kw_sequence = &IdentTable.get("sequence");
1120     kw_small = &IdentTable.get("small");
1121     kw_soft = &IdentTable.get("soft");
1122     kw_solve = &IdentTable.get("solve");
1123     kw_specify = &IdentTable.get("specify");
1124     kw_specparam = &IdentTable.get("specparam");
1125     kw_strong0 = &IdentTable.get("strong0");
1126     kw_strong1 = &IdentTable.get("strong1");
1127     kw_supply0 = &IdentTable.get("supply0");
1128     kw_supply1 = &IdentTable.get("supply1");
1129     kw_table = &IdentTable.get("table");
1130     kw_tagged = &IdentTable.get("tagged");
1131     kw_task = &IdentTable.get("task");
1132     kw_timescale = &IdentTable.get("timescale");
1133     kw_tri = &IdentTable.get("tri");
1134     kw_tri0 = &IdentTable.get("tri0");
1135     kw_tri1 = &IdentTable.get("tri1");
1136     kw_triand = &IdentTable.get("triand");
1137     kw_trior = &IdentTable.get("trior");
1138     kw_trireg = &IdentTable.get("trireg");
1139     kw_unconnected_drive = &IdentTable.get("unconnected_drive");
1140     kw_undefineall = &IdentTable.get("undefineall");
1141     kw_unique = &IdentTable.get("unique");
1142     kw_unique0 = &IdentTable.get("unique0");
1143     kw_uwire = &IdentTable.get("uwire");
1144     kw_vectored = &IdentTable.get("vectored");
1145     kw_wand = &IdentTable.get("wand");
1146     kw_weak0 = &IdentTable.get("weak0");
1147     kw_weak1 = &IdentTable.get("weak1");
1148     kw_wildcard = &IdentTable.get("wildcard");
1149     kw_wire = &IdentTable.get("wire");
1150     kw_with = &IdentTable.get("with");
1151     kw_wor = &IdentTable.get("wor");
1152 
1153     // Symbols that are treated as keywords.
1154     kw_verilogHash = &IdentTable.get("#");
1155     kw_verilogHashHash = &IdentTable.get("##");
1156     kw_apostrophe = &IdentTable.get("\'");
1157 
1158     // Keep this at the end of the constructor to make sure everything here
1159     // is
1160     // already initialized.
1161     JsExtraKeywords = std::unordered_set<IdentifierInfo *>(
1162         {kw_as, kw_async, kw_await, kw_declare, kw_finally, kw_from,
1163          kw_function, kw_get, kw_import, kw_is, kw_let, kw_module, kw_override,
1164          kw_readonly, kw_set, kw_type, kw_typeof, kw_var, kw_yield,
1165          // Keywords from the Java section.
1166          kw_abstract, kw_extends, kw_implements, kw_instanceof, kw_interface});
1167 
1168     CSharpExtraKeywords = std::unordered_set<IdentifierInfo *>(
1169         {kw_base, kw_byte, kw_checked, kw_decimal, kw_delegate, kw_event,
1170          kw_fixed, kw_foreach, kw_implicit, kw_in, kw_init, kw_interface,
1171          kw_internal, kw_is, kw_lock, kw_null, kw_object, kw_out, kw_override,
1172          kw_params, kw_readonly, kw_ref, kw_string, kw_stackalloc, kw_sbyte,
1173          kw_sealed, kw_uint, kw_ulong, kw_unchecked, kw_unsafe, kw_ushort,
1174          kw_when, kw_where,
1175          // Keywords from the JavaScript section.
1176          kw_as, kw_async, kw_await, kw_declare, kw_finally, kw_from,
1177          kw_function, kw_get, kw_import, kw_is, kw_let, kw_module, kw_readonly,
1178          kw_set, kw_type, kw_typeof, kw_var, kw_yield,
1179          // Keywords from the Java section.
1180          kw_abstract, kw_extends, kw_implements, kw_instanceof, kw_interface});
1181 
1182     // Some keywords are not included here because they don't need special
1183     // treatment like `showcancelled` or they should be treated as identifiers
1184     // like `int` and `logic`.
1185     VerilogExtraKeywords =
1186         std::unordered_set<IdentifierInfo *>({kw_always,
1187                                               kw_always_comb,
1188                                               kw_always_ff,
1189                                               kw_always_latch,
1190                                               kw_assert,
1191                                               kw_assign,
1192                                               kw_assume,
1193                                               kw_automatic,
1194                                               kw_before,
1195                                               kw_begin,
1196                                               kw_bins,
1197                                               kw_binsof,
1198                                               kw_casex,
1199                                               kw_casez,
1200                                               kw_celldefine,
1201                                               kw_checker,
1202                                               kw_clocking,
1203                                               kw_constraint,
1204                                               kw_cover,
1205                                               kw_covergroup,
1206                                               kw_coverpoint,
1207                                               kw_disable,
1208                                               kw_dist,
1209                                               kw_end,
1210                                               kw_endcase,
1211                                               kw_endchecker,
1212                                               kw_endclass,
1213                                               kw_endclocking,
1214                                               kw_endfunction,
1215                                               kw_endgenerate,
1216                                               kw_endgroup,
1217                                               kw_endinterface,
1218                                               kw_endmodule,
1219                                               kw_endpackage,
1220                                               kw_endprimitive,
1221                                               kw_endprogram,
1222                                               kw_endproperty,
1223                                               kw_endsequence,
1224                                               kw_endspecify,
1225                                               kw_endtable,
1226                                               kw_endtask,
1227                                               kw_extends,
1228                                               kw_final,
1229                                               kw_foreach,
1230                                               kw_forever,
1231                                               kw_fork,
1232                                               kw_function,
1233                                               kw_generate,
1234                                               kw_highz0,
1235                                               kw_highz1,
1236                                               kw_iff,
1237                                               kw_ifnone,
1238                                               kw_ignore_bins,
1239                                               kw_illegal_bins,
1240                                               kw_implements,
1241                                               kw_import,
1242                                               kw_initial,
1243                                               kw_inout,
1244                                               kw_input,
1245                                               kw_inside,
1246                                               kw_interconnect,
1247                                               kw_interface,
1248                                               kw_intersect,
1249                                               kw_join,
1250                                               kw_join_any,
1251                                               kw_join_none,
1252                                               kw_large,
1253                                               kw_let,
1254                                               kw_local,
1255                                               kw_localparam,
1256                                               kw_macromodule,
1257                                               kw_matches,
1258                                               kw_medium,
1259                                               kw_output,
1260                                               kw_package,
1261                                               kw_packed,
1262                                               kw_parameter,
1263                                               kw_primitive,
1264                                               kw_priority,
1265                                               kw_program,
1266                                               kw_property,
1267                                               kw_pull0,
1268                                               kw_pull1,
1269                                               kw_pure,
1270                                               kw_rand,
1271                                               kw_randc,
1272                                               kw_randcase,
1273                                               kw_randsequence,
1274                                               kw_ref,
1275                                               kw_repeat,
1276                                               kw_sample,
1277                                               kw_scalared,
1278                                               kw_sequence,
1279                                               kw_small,
1280                                               kw_soft,
1281                                               kw_solve,
1282                                               kw_specify,
1283                                               kw_specparam,
1284                                               kw_strong0,
1285                                               kw_strong1,
1286                                               kw_supply0,
1287                                               kw_supply1,
1288                                               kw_table,
1289                                               kw_tagged,
1290                                               kw_task,
1291                                               kw_tri,
1292                                               kw_tri0,
1293                                               kw_tri1,
1294                                               kw_triand,
1295                                               kw_trior,
1296                                               kw_trireg,
1297                                               kw_unique,
1298                                               kw_unique0,
1299                                               kw_uwire,
1300                                               kw_var,
1301                                               kw_vectored,
1302                                               kw_wand,
1303                                               kw_weak0,
1304                                               kw_weak1,
1305                                               kw_wildcard,
1306                                               kw_wire,
1307                                               kw_with,
1308                                               kw_wor,
1309                                               kw_verilogHash,
1310                                               kw_verilogHashHash});
1311   }
1312 
1313   // Context sensitive keywords.
1314   IdentifierInfo *kw_final;
1315   IdentifierInfo *kw_override;
1316   IdentifierInfo *kw_in;
1317   IdentifierInfo *kw_of;
1318   IdentifierInfo *kw_CF_CLOSED_ENUM;
1319   IdentifierInfo *kw_CF_ENUM;
1320   IdentifierInfo *kw_CF_OPTIONS;
1321   IdentifierInfo *kw_NS_CLOSED_ENUM;
1322   IdentifierInfo *kw_NS_ENUM;
1323   IdentifierInfo *kw_NS_OPTIONS;
1324   IdentifierInfo *kw___except;
1325   IdentifierInfo *kw___has_include;
1326   IdentifierInfo *kw___has_include_next;
1327 
1328   // JavaScript keywords.
1329   IdentifierInfo *kw_as;
1330   IdentifierInfo *kw_async;
1331   IdentifierInfo *kw_await;
1332   IdentifierInfo *kw_declare;
1333   IdentifierInfo *kw_finally;
1334   IdentifierInfo *kw_from;
1335   IdentifierInfo *kw_function;
1336   IdentifierInfo *kw_get;
1337   IdentifierInfo *kw_import;
1338   IdentifierInfo *kw_infer;
1339   IdentifierInfo *kw_is;
1340   IdentifierInfo *kw_let;
1341   IdentifierInfo *kw_module;
1342   IdentifierInfo *kw_readonly;
1343   IdentifierInfo *kw_set;
1344   IdentifierInfo *kw_type;
1345   IdentifierInfo *kw_typeof;
1346   IdentifierInfo *kw_var;
1347   IdentifierInfo *kw_yield;
1348 
1349   // Java keywords.
1350   IdentifierInfo *kw_abstract;
1351   IdentifierInfo *kw_assert;
1352   IdentifierInfo *kw_extends;
1353   IdentifierInfo *kw_implements;
1354   IdentifierInfo *kw_instanceof;
1355   IdentifierInfo *kw_interface;
1356   IdentifierInfo *kw_native;
1357   IdentifierInfo *kw_package;
1358   IdentifierInfo *kw_synchronized;
1359   IdentifierInfo *kw_throws;
1360 
1361   // Pragma keywords.
1362   IdentifierInfo *kw_mark;
1363   IdentifierInfo *kw_region;
1364 
1365   // Proto keywords.
1366   IdentifierInfo *kw_extend;
1367   IdentifierInfo *kw_option;
1368   IdentifierInfo *kw_optional;
1369   IdentifierInfo *kw_repeated;
1370   IdentifierInfo *kw_required;
1371   IdentifierInfo *kw_returns;
1372 
1373   // QT keywords.
1374   IdentifierInfo *kw_signals;
1375   IdentifierInfo *kw_qsignals;
1376   IdentifierInfo *kw_slots;
1377   IdentifierInfo *kw_qslots;
1378 
1379   // For internal use by clang-format.
1380   IdentifierInfo *kw_internal_ident_after_define;
1381 
1382   // C# keywords
1383   IdentifierInfo *kw_dollar;
1384   IdentifierInfo *kw_base;
1385   IdentifierInfo *kw_byte;
1386   IdentifierInfo *kw_checked;
1387   IdentifierInfo *kw_decimal;
1388   IdentifierInfo *kw_delegate;
1389   IdentifierInfo *kw_event;
1390   IdentifierInfo *kw_fixed;
1391   IdentifierInfo *kw_foreach;
1392   IdentifierInfo *kw_implicit;
1393   IdentifierInfo *kw_init;
1394   IdentifierInfo *kw_internal;
1395 
1396   IdentifierInfo *kw_lock;
1397   IdentifierInfo *kw_null;
1398   IdentifierInfo *kw_object;
1399   IdentifierInfo *kw_out;
1400 
1401   IdentifierInfo *kw_params;
1402 
1403   IdentifierInfo *kw_ref;
1404   IdentifierInfo *kw_string;
1405   IdentifierInfo *kw_stackalloc;
1406   IdentifierInfo *kw_sbyte;
1407   IdentifierInfo *kw_sealed;
1408   IdentifierInfo *kw_uint;
1409   IdentifierInfo *kw_ulong;
1410   IdentifierInfo *kw_unchecked;
1411   IdentifierInfo *kw_unsafe;
1412   IdentifierInfo *kw_ushort;
1413   IdentifierInfo *kw_when;
1414   IdentifierInfo *kw_where;
1415 
1416   // Verilog keywords
1417   IdentifierInfo *kw_always;
1418   IdentifierInfo *kw_always_comb;
1419   IdentifierInfo *kw_always_ff;
1420   IdentifierInfo *kw_always_latch;
1421   IdentifierInfo *kw_assign;
1422   IdentifierInfo *kw_assume;
1423   IdentifierInfo *kw_automatic;
1424   IdentifierInfo *kw_before;
1425   IdentifierInfo *kw_begin;
1426   IdentifierInfo *kw_begin_keywords;
1427   IdentifierInfo *kw_bins;
1428   IdentifierInfo *kw_binsof;
1429   IdentifierInfo *kw_casex;
1430   IdentifierInfo *kw_casez;
1431   IdentifierInfo *kw_celldefine;
1432   IdentifierInfo *kw_checker;
1433   IdentifierInfo *kw_clocking;
1434   IdentifierInfo *kw_constraint;
1435   IdentifierInfo *kw_cover;
1436   IdentifierInfo *kw_covergroup;
1437   IdentifierInfo *kw_coverpoint;
1438   IdentifierInfo *kw_default_decay_time;
1439   IdentifierInfo *kw_default_nettype;
1440   IdentifierInfo *kw_default_trireg_strength;
1441   IdentifierInfo *kw_delay_mode_distributed;
1442   IdentifierInfo *kw_delay_mode_path;
1443   IdentifierInfo *kw_delay_mode_unit;
1444   IdentifierInfo *kw_delay_mode_zero;
1445   IdentifierInfo *kw_disable;
1446   IdentifierInfo *kw_dist;
1447   IdentifierInfo *kw_elsif;
1448   IdentifierInfo *kw_end;
1449   IdentifierInfo *kw_end_keywords;
1450   IdentifierInfo *kw_endcase;
1451   IdentifierInfo *kw_endcelldefine;
1452   IdentifierInfo *kw_endchecker;
1453   IdentifierInfo *kw_endclass;
1454   IdentifierInfo *kw_endclocking;
1455   IdentifierInfo *kw_endfunction;
1456   IdentifierInfo *kw_endgenerate;
1457   IdentifierInfo *kw_endgroup;
1458   IdentifierInfo *kw_endinterface;
1459   IdentifierInfo *kw_endmodule;
1460   IdentifierInfo *kw_endpackage;
1461   IdentifierInfo *kw_endprimitive;
1462   IdentifierInfo *kw_endprogram;
1463   IdentifierInfo *kw_endproperty;
1464   IdentifierInfo *kw_endsequence;
1465   IdentifierInfo *kw_endspecify;
1466   IdentifierInfo *kw_endtable;
1467   IdentifierInfo *kw_endtask;
1468   IdentifierInfo *kw_forever;
1469   IdentifierInfo *kw_fork;
1470   IdentifierInfo *kw_generate;
1471   IdentifierInfo *kw_highz0;
1472   IdentifierInfo *kw_highz1;
1473   IdentifierInfo *kw_iff;
1474   IdentifierInfo *kw_ifnone;
1475   IdentifierInfo *kw_ignore_bins;
1476   IdentifierInfo *kw_illegal_bins;
1477   IdentifierInfo *kw_initial;
1478   IdentifierInfo *kw_inout;
1479   IdentifierInfo *kw_input;
1480   IdentifierInfo *kw_inside;
1481   IdentifierInfo *kw_interconnect;
1482   IdentifierInfo *kw_intersect;
1483   IdentifierInfo *kw_join;
1484   IdentifierInfo *kw_join_any;
1485   IdentifierInfo *kw_join_none;
1486   IdentifierInfo *kw_large;
1487   IdentifierInfo *kw_local;
1488   IdentifierInfo *kw_localparam;
1489   IdentifierInfo *kw_macromodule;
1490   IdentifierInfo *kw_matches;
1491   IdentifierInfo *kw_medium;
1492   IdentifierInfo *kw_nounconnected_drive;
1493   IdentifierInfo *kw_output;
1494   IdentifierInfo *kw_packed;
1495   IdentifierInfo *kw_parameter;
1496   IdentifierInfo *kw_primitive;
1497   IdentifierInfo *kw_priority;
1498   IdentifierInfo *kw_program;
1499   IdentifierInfo *kw_property;
1500   IdentifierInfo *kw_pull0;
1501   IdentifierInfo *kw_pull1;
1502   IdentifierInfo *kw_pure;
1503   IdentifierInfo *kw_rand;
1504   IdentifierInfo *kw_randc;
1505   IdentifierInfo *kw_randcase;
1506   IdentifierInfo *kw_randsequence;
1507   IdentifierInfo *kw_repeat;
1508   IdentifierInfo *kw_resetall;
1509   IdentifierInfo *kw_sample;
1510   IdentifierInfo *kw_scalared;
1511   IdentifierInfo *kw_sequence;
1512   IdentifierInfo *kw_small;
1513   IdentifierInfo *kw_soft;
1514   IdentifierInfo *kw_solve;
1515   IdentifierInfo *kw_specify;
1516   IdentifierInfo *kw_specparam;
1517   IdentifierInfo *kw_strong0;
1518   IdentifierInfo *kw_strong1;
1519   IdentifierInfo *kw_supply0;
1520   IdentifierInfo *kw_supply1;
1521   IdentifierInfo *kw_table;
1522   IdentifierInfo *kw_tagged;
1523   IdentifierInfo *kw_task;
1524   IdentifierInfo *kw_timescale;
1525   IdentifierInfo *kw_tri0;
1526   IdentifierInfo *kw_tri1;
1527   IdentifierInfo *kw_tri;
1528   IdentifierInfo *kw_triand;
1529   IdentifierInfo *kw_trior;
1530   IdentifierInfo *kw_trireg;
1531   IdentifierInfo *kw_unconnected_drive;
1532   IdentifierInfo *kw_undefineall;
1533   IdentifierInfo *kw_unique;
1534   IdentifierInfo *kw_unique0;
1535   IdentifierInfo *kw_uwire;
1536   IdentifierInfo *kw_vectored;
1537   IdentifierInfo *kw_wand;
1538   IdentifierInfo *kw_weak0;
1539   IdentifierInfo *kw_weak1;
1540   IdentifierInfo *kw_wildcard;
1541   IdentifierInfo *kw_wire;
1542   IdentifierInfo *kw_with;
1543   IdentifierInfo *kw_wor;
1544 
1545   // Workaround for hashes and backticks in Verilog.
1546   IdentifierInfo *kw_verilogHash;
1547   IdentifierInfo *kw_verilogHashHash;
1548 
1549   // Symbols in Verilog that don't exist in C++.
1550   IdentifierInfo *kw_apostrophe;
1551 
1552   /// Returns \c true if \p Tok is a keyword or an identifier.
isWordLikeAdditionalKeywords1553   bool isWordLike(const FormatToken &Tok) const {
1554     // getIdentifierinfo returns non-null for keywords as well as identifiers.
1555     return Tok.Tok.getIdentifierInfo() != nullptr &&
1556            !Tok.isOneOf(kw_verilogHash, kw_verilogHashHash, kw_apostrophe);
1557   }
1558 
1559   /// Returns \c true if \p Tok is a true JavaScript identifier, returns
1560   /// \c false if it is a keyword or a pseudo keyword.
1561   /// If \c AcceptIdentifierName is true, returns true not only for keywords,
1562   // but also for IdentifierName tokens (aka pseudo-keywords), such as
1563   // ``yield``.
1564   bool IsJavaScriptIdentifier(const FormatToken &Tok,
1565                               bool AcceptIdentifierName = true) const {
1566     // Based on the list of JavaScript & TypeScript keywords here:
1567     // https://github.com/microsoft/TypeScript/blob/main/src/compiler/scanner.ts#L74
1568     switch (Tok.Tok.getKind()) {
1569     case tok::kw_break:
1570     case tok::kw_case:
1571     case tok::kw_catch:
1572     case tok::kw_class:
1573     case tok::kw_continue:
1574     case tok::kw_const:
1575     case tok::kw_default:
1576     case tok::kw_delete:
1577     case tok::kw_do:
1578     case tok::kw_else:
1579     case tok::kw_enum:
1580     case tok::kw_export:
1581     case tok::kw_false:
1582     case tok::kw_for:
1583     case tok::kw_if:
1584     case tok::kw_import:
1585     case tok::kw_module:
1586     case tok::kw_new:
1587     case tok::kw_private:
1588     case tok::kw_protected:
1589     case tok::kw_public:
1590     case tok::kw_return:
1591     case tok::kw_static:
1592     case tok::kw_switch:
1593     case tok::kw_this:
1594     case tok::kw_throw:
1595     case tok::kw_true:
1596     case tok::kw_try:
1597     case tok::kw_typeof:
1598     case tok::kw_void:
1599     case tok::kw_while:
1600       // These are JS keywords that are lexed by LLVM/clang as keywords.
1601       return false;
1602     case tok::identifier: {
1603       // For identifiers, make sure they are true identifiers, excluding the
1604       // JavaScript pseudo-keywords (not lexed by LLVM/clang as keywords).
1605       bool IsPseudoKeyword =
1606           JsExtraKeywords.find(Tok.Tok.getIdentifierInfo()) !=
1607           JsExtraKeywords.end();
1608       return AcceptIdentifierName || !IsPseudoKeyword;
1609     }
1610     default:
1611       // Other keywords are handled in the switch below, to avoid problems due
1612       // to duplicate case labels when using the #include trick.
1613       break;
1614     }
1615 
1616     switch (Tok.Tok.getKind()) {
1617       // Handle C++ keywords not included above: these are all JS identifiers.
1618 #define KEYWORD(X, Y) case tok::kw_##X:
1619 #include "clang/Basic/TokenKinds.def"
1620       // #undef KEYWORD is not needed -- it's #undef-ed at the end of
1621       // TokenKinds.def
1622       return true;
1623     default:
1624       // All other tokens (punctuation etc) are not JS identifiers.
1625       return false;
1626     }
1627   }
1628 
1629   /// Returns \c true if \p Tok is a C# keyword, returns
1630   /// \c false if it is a anything else.
isCSharpKeywordAdditionalKeywords1631   bool isCSharpKeyword(const FormatToken &Tok) const {
1632     switch (Tok.Tok.getKind()) {
1633     case tok::kw_bool:
1634     case tok::kw_break:
1635     case tok::kw_case:
1636     case tok::kw_catch:
1637     case tok::kw_char:
1638     case tok::kw_class:
1639     case tok::kw_const:
1640     case tok::kw_continue:
1641     case tok::kw_default:
1642     case tok::kw_do:
1643     case tok::kw_double:
1644     case tok::kw_else:
1645     case tok::kw_enum:
1646     case tok::kw_explicit:
1647     case tok::kw_extern:
1648     case tok::kw_false:
1649     case tok::kw_float:
1650     case tok::kw_for:
1651     case tok::kw_goto:
1652     case tok::kw_if:
1653     case tok::kw_int:
1654     case tok::kw_long:
1655     case tok::kw_namespace:
1656     case tok::kw_new:
1657     case tok::kw_operator:
1658     case tok::kw_private:
1659     case tok::kw_protected:
1660     case tok::kw_public:
1661     case tok::kw_return:
1662     case tok::kw_short:
1663     case tok::kw_sizeof:
1664     case tok::kw_static:
1665     case tok::kw_struct:
1666     case tok::kw_switch:
1667     case tok::kw_this:
1668     case tok::kw_throw:
1669     case tok::kw_true:
1670     case tok::kw_try:
1671     case tok::kw_typeof:
1672     case tok::kw_using:
1673     case tok::kw_virtual:
1674     case tok::kw_void:
1675     case tok::kw_volatile:
1676     case tok::kw_while:
1677       return true;
1678     default:
1679       return Tok.is(tok::identifier) &&
1680              CSharpExtraKeywords.find(Tok.Tok.getIdentifierInfo()) ==
1681                  CSharpExtraKeywords.end();
1682     }
1683   }
1684 
isVerilogWordOperatorAdditionalKeywords1685   bool isVerilogWordOperator(const FormatToken &Tok) const {
1686     return Tok.isOneOf(kw_before, kw_intersect, kw_dist, kw_iff, kw_inside,
1687                        kw_with);
1688   }
1689 
isVerilogIdentifierAdditionalKeywords1690   bool isVerilogIdentifier(const FormatToken &Tok) const {
1691     switch (Tok.Tok.getKind()) {
1692     case tok::kw_case:
1693     case tok::kw_class:
1694     case tok::kw_const:
1695     case tok::kw_continue:
1696     case tok::kw_default:
1697     case tok::kw_do:
1698     case tok::kw_extern:
1699     case tok::kw_else:
1700     case tok::kw_enum:
1701     case tok::kw_for:
1702     case tok::kw_if:
1703     case tok::kw_restrict:
1704     case tok::kw_signed:
1705     case tok::kw_static:
1706     case tok::kw_struct:
1707     case tok::kw_typedef:
1708     case tok::kw_union:
1709     case tok::kw_unsigned:
1710     case tok::kw_virtual:
1711     case tok::kw_while:
1712       return false;
1713     case tok::identifier:
1714       return VerilogExtraKeywords.find(Tok.Tok.getIdentifierInfo()) ==
1715              VerilogExtraKeywords.end();
1716     default:
1717       // getIdentifierInfo returns non-null for both identifiers and keywords.
1718       return Tok.Tok.getIdentifierInfo() != nullptr;
1719     }
1720   }
1721 
1722   /// Returns whether \p Tok is a Verilog preprocessor directive.  This is
1723   /// needed because macro expansions start with a backtick as well and they
1724   /// need to be treated differently.
isVerilogPPDirectiveAdditionalKeywords1725   bool isVerilogPPDirective(const FormatToken &Tok) const {
1726     auto Info = Tok.Tok.getIdentifierInfo();
1727     if (!Info)
1728       return false;
1729     switch (Info->getPPKeywordID()) {
1730     case tok::pp_define:
1731     case tok::pp_else:
1732     case tok::pp_endif:
1733     case tok::pp_ifdef:
1734     case tok::pp_ifndef:
1735     case tok::pp_include:
1736     case tok::pp_line:
1737     case tok::pp_pragma:
1738     case tok::pp_undef:
1739       return true;
1740     default:
1741       return Tok.isOneOf(kw_begin_keywords, kw_celldefine,
1742                          kw_default_decay_time, kw_default_nettype,
1743                          kw_default_trireg_strength, kw_delay_mode_distributed,
1744                          kw_delay_mode_path, kw_delay_mode_unit,
1745                          kw_delay_mode_zero, kw_elsif, kw_end_keywords,
1746                          kw_endcelldefine, kw_nounconnected_drive, kw_resetall,
1747                          kw_timescale, kw_unconnected_drive, kw_undefineall);
1748     }
1749   }
1750 
1751   /// Returns whether \p Tok is a Verilog keyword that opens a block.
isVerilogBeginAdditionalKeywords1752   bool isVerilogBegin(const FormatToken &Tok) const {
1753     // `table` is not included since it needs to be treated specially.
1754     return !Tok.endsSequence(kw_fork, kw_disable) &&
1755            Tok.isOneOf(kw_begin, kw_fork, kw_generate, kw_specify);
1756   }
1757 
1758   /// Returns whether \p Tok is a Verilog keyword that closes a block.
isVerilogEndAdditionalKeywords1759   bool isVerilogEnd(const FormatToken &Tok) const {
1760     return !Tok.endsSequence(kw_join, kw_rand) &&
1761            Tok.isOneOf(TT_MacroBlockEnd, kw_end, kw_endcase, kw_endclass,
1762                        kw_endclocking, kw_endchecker, kw_endfunction,
1763                        kw_endgenerate, kw_endgroup, kw_endinterface,
1764                        kw_endmodule, kw_endpackage, kw_endprimitive,
1765                        kw_endprogram, kw_endproperty, kw_endsequence,
1766                        kw_endspecify, kw_endtable, kw_endtask, kw_join,
1767                        kw_join_any, kw_join_none);
1768   }
1769 
1770   /// Returns whether \p Tok is a Verilog keyword that opens a module, etc.
isVerilogHierarchyAdditionalKeywords1771   bool isVerilogHierarchy(const FormatToken &Tok) const {
1772     if (Tok.endsSequence(kw_function, kw_with))
1773       return false;
1774     if (Tok.is(kw_property)) {
1775       const FormatToken *Prev = Tok.getPreviousNonComment();
1776       return !(Prev &&
1777                Prev->isOneOf(tok::kw_restrict, kw_assert, kw_assume, kw_cover));
1778     }
1779     return Tok.isOneOf(tok::kw_case, tok::kw_class, kw_function, kw_module,
1780                        kw_interface, kw_package, kw_casex, kw_casez, kw_checker,
1781                        kw_clocking, kw_covergroup, kw_macromodule, kw_primitive,
1782                        kw_program, kw_property, kw_randcase, kw_randsequence,
1783                        kw_task);
1784   }
1785 
isVerilogEndOfLabelAdditionalKeywords1786   bool isVerilogEndOfLabel(const FormatToken &Tok) const {
1787     const FormatToken *Next = Tok.getNextNonComment();
1788     // In Verilog the colon in a default label is optional.
1789     return Tok.is(TT_GotoLabelColon) ||
1790            (Tok.is(tok::kw_default) &&
1791             !(Next && Next->isOneOf(tok::colon, tok::semi, kw_clocking, kw_iff,
1792                                     kw_input, kw_output, kw_sequence)));
1793   }
1794 
1795 private:
1796   /// The JavaScript keywords beyond the C++ keyword set.
1797   std::unordered_set<IdentifierInfo *> JsExtraKeywords;
1798 
1799   /// The C# keywords beyond the C++ keyword set
1800   std::unordered_set<IdentifierInfo *> CSharpExtraKeywords;
1801 
1802   /// The Verilog keywords beyond the C++ keyword set.
1803   std::unordered_set<IdentifierInfo *> VerilogExtraKeywords;
1804 };
1805 
1806 } // namespace format
1807 } // namespace clang
1808 
1809 #endif
1810